AI companies are seeing revenue growth in multiples of compute growth. Is this sustainable?
AI companies are seeing revenue growth in multiples of compute growth. Is this sustainable?
Description
AI companies are achieving exponential revenue growth despite hardware capacity constraints limiting their available compute
Transcript
is it that they're only able to increase their compute, so effectively their capacity to deliver AI by 3x but their revenue is able to go up by 10x? What that is as a data point, right, so when they go in there they've only got, you know, an extra few gigawatts of power to be able to deliver, or compute with powered compute to be able to deliver the AI, and yet their revenue is going one billion to ten billion and it looks like it's about to go ten billion to one hundred billion next year. And they were already running at max capacity, exactly. So it's not enough, maybe there's just a lot more like, exactly, a lot more usage. So with that in, with those things that we know to be true as facts, you've got a very hard data point there that so far being able to increase the price has not been a problem, and then there's a question underneath that of how much of that increase in price are they retaining as their own margin versus how much are they just having to pay to the actual computer providers, so the providers of data centers and chips.

